This forum post has been removed
Number of replies: 2
The content of this forum post has been removed and can no longer be accessed.
I feel your pain. This is the last piece of the puzzle for us.
If you're using a SAN, you might be able to offload the HA to that layer. NetApp filers can be configured to present NFS4 shares, and to have a second point of presence in another datacentre. I think this isn't true HA, though, rather the NetApp will take snapshots from time to time, so there is still the risk of data loss.
I know people here have used glusterfs. Ceph has had POSIX filesystem semantics for a while now, and looks very futuristic, but I haven't had the guts to use it in production.
Our organization uses a shared LUN set up with replication going between two data centers. Check out GPFS or GFS. We get local disk speeds on these shared mounts and we have yet to have a problem with them after we went live. Just expect to have to increase the amount of RAM available on your web nodes because these luns need a lot of page file space when performing write operations.